I've gathered few measurements about couple of heatsinks to this picture. Dynatron data is supplied by manufacturer, and all other data is taken from spcr tests. So if you don't believe what you see, there might be a good reason for that Dynatron K618 vs. K520 shows dramatic difference between std aluminium HS and heatpipe HS when CFM is extremely slow.

I've build two near silent HTPC:s now, and both of them used aluminium heatsinks with 80mm NF-R8 Noctua fan. I've got Scythe Ninja Mini too, and it was completely useless when I tried to build silent rig. My excel shows why My first silent htpc setup used intel oem 274g aluminium heatsink for 2.4GHz P4. 80mm Noctua (with U.L.N.A) kept the system way more cooler that with Scythe (same fan, same rpm).

For my current htpc I didn't even thought of putting heatpipe cooler according to my previous experiences. So I put dynatron K520 for i3-2100T (CPU is my only GPU too). Fan is still the same 80mm Noctua with U.L.N.A. Normal htpc use rises CPU temperature typically not more than 5 degrees above case ambient. I've got only one case fan, which is installed inside the PSU. Fan is old Noctua 120mm (don't remember the model) controlled by PSU. Case temperature is typically 10 degrees above ambient temperature. From two meters (sofa) it's impossible to hear my htpc.

I don't understand how you put together airflow and the use of heatpipes. Makes no sense to me.

Heatpipes are just a nice way to transfer heat. If they are useful (and cost effective) depends on the construction of the cooler. In my experience the Ninja Mini, with its wide spaced fins, is excellent for low airflow cooling. If that other cooler had tighter fin spacing I'm not surprised that it performs better, given a decent airflow.

I cool my i3 2100T with an old Dell Optiplex cooler I found in the electronics recycle bin. Zip tied to the CPU, no fan (it gets a little airflow from the case fan). You can cool these chips with very little. They can also handle a bit of heat, no need to struggle to keep the chip close to ambient imo.

Look Dynatron K520 vs. K618 mentioned earlier. You can see from the picture that under 22-23 CFM K618 heatpipe version is useless compared to K520. 22 CFM corresponds 1500-2000rpm 80mm fan or ~1000rpm 120mm fan. Both too loud for silent htpc - no matter what the fans are.

Or if you compare Scythe Ninja mini to K520 you see that K520 is more efficient if air flow rate is less than 18 CFM. This i can easily believe as I've tried Ninja mini with 800rpm (~10 CFM) 80mm fan. As I already mentioned P4 oem aluminium HS was way more efficient in the same system with the same fans (=CFM).

Of course K520 (or even smaller aluminium heatsink) can easily handle i3-2100T without a CPU fan. I guess that it's enough to have around 1 °C / W cooling for i3-2100T. From the picture you can estimate that it's achieved with a very little airflow by any HS.

What about static pressure surely a pretty important factor here?
Again I'm more interested in real world performance rather than manufacturers spec sheets I'm not convinced most of them are very accurate!

Of course there are many other factors, and spcr tests cannot be compared to Dynatron tunnel tests. Thats why there is a question mark in the subject title. I wish spcr to test few good heatpipe coolers and one basic heatsink like Dynatron K520 with wider range of air flow (lets say 5-40 CFM).

I find that hard to believe.

The CPU you're using uses so little power that the HSF is pretty much irrelevant. The only time you need a heat-pipe HSF is with a CPU that is drawing more (or way more) than 60W at load.

BTW, many of the heat sinks in your comparison are junk, especially the Ninja Mini. The best heat pipe HSFs to use as a base are the Megahalems and HR-01, two of the few top-of-the-line heat sinks still available. Of course, they show up as the best in your graph above...

I remember reading something about heatpipe coolers not bieng soo good a low temps, but nothing related directly to CFMs.

Here you can see few days (24/7 on) data of my current htpc with 'normal' use. I don't play games. Ambient temperature has been around 23-25. I've noticed that 7" touchscreen on my Cooltek 3000 case rises temperature quite a lot.

I still have one P4 2.4GHz running, so I can test again the difference between oem hs and Ninja Mini.

I've learned now more theory about heat pipes, and it seems obvious that their efficiency rises when temperature difference rises between CPU and heatsinks. That's why they don't work so well in a low power quiet systems.

My experience is quite different. I've been "passively" cooling CPUs for years ( by that I mean that there is no fan on the actual CPU cooler, but I always have a fan on the case ). My current i7 2600k is cooled by a HR-02, without any fan on it. The load temps I get are 77C ( using IBT 2.5.1, for 6 hours, with a 33C ambient and no AC; typical temps after 3-4 hours of gaming are in the high 50s - 56-58 ). My wife's rig ( e6850 cpu ) is cooled by a Scythe Ninja ( rev 1 ) with a 500 RPM Scythe on the case. Her load temps are somewhere in the high 50s ( 58 or so ). In your case I guess that the very low TDP of your CPU doesn't really require a better cooler. Also, as far as I can remember the installation of the Scythe Ninja Mini is done with push-pins ( pretty much the main reason I've given up on Ninja's and went to TR ), which is definitely not optimum for a heavy heatsink.

I still don't get it. Dynatron K520 vs. K618.. They are two different coolers. It's not as the heatpipes are the only (or most significant) difference between the two. Take a look at the fin spacing. No wonder the K520 does better at low airflow.
Also, the numbers are from different sources, some are numbers provided by the manufacturers, some are from SPCR tests. There is no way to compare them. You have to test them both using the same method to do that.


The Minja is not junk. It's a great cooler for a low airflow setup where a bigger cooler wouldn't fit.

Fair enough; it's an okay miniature cooler, but its mounting system is poor. I should perhaps have said "uncompetitive" rather than "junk".

I didn't express myself clear enough then. How do you understand this? I even gave a hint of that in the original message:

So you say that heatpipes has nothing to do with the poor efficiency at low CFM? It's about fin spacing then? In real life it's much more complex as you can see if you read this wikipedia article.

Dynatron tests are made in "wind tunnel". If all the air is forced through cooler, then K618 must be even worse in real life at low CFM.

I still haven't seen any test results - which makes me wonder if anyone of you have even tested any basic heatsink without heatpipes?

I'm not saying that you are dead wrong, I'm saying that you draw conclusions that I don't agree with. I'd say for example that I'd rather think that fin spacing has more impact on low airflow performance than heatpipes, at least when comparing the two Dynatron coolers. At least those two we can compare since the data comes from a single source.

You say that "spcr tests cannot be compared to Dynatron tunnel tests" but still you say "if you compare Scythe Ninja mini to K520 you see that K520 is more efficient if air flow rate is less than 18 CFM". Again, makes no sense to me.

You get worse results from a 580g Ninja Mini than with an Intel OEM 274g aluminium heatsink and gets defensive when people are doubtful? You need to back up your findings with more than anecdotes.


Yeah, it's a rather old cooler today and the mounting system has always been Scythes Achilles heel. I've never had to use the crappy push-pins though, the AM2 mounting system is better.

I think this sounds right.
Heatpipes don't work as well (don't transfer heat) when heatsink fin temperature is high. I think I've seen tests where they overload heatpipe heat transfer capability and heatsink temperature skyrockets. It work other way round too.
My internal explanation would be that the vapour inside the pipe doesn't condense on the "cool" part anymore.

With regular aluminum heatsink no such problem, speed of heat transfer only depends on temperature difference between base and fins, regardles of actual values.

I tried to bring clarity to topic but afraid messed it up even more with this post, sorry

I'd rather see some real test results. Tighter fin spacing means in this case (K618 vs. K520) also more surface area in the cooler. In theory that should help cooling no matter what the forced air flow rate is.



Sorry about that. I just tried to make you understand the picture.

Not always; the original Ninja (SCNJ-1000) had both an excellent if somewhat clumsy mounting system (hooks and rails) and unusually good base/heatpipe contact (soldered). Heck there was even solder between the heat pipes and the fins. One of the few best heat sinks ever.